Brittany Trafis runs an AI search company, and she can trace one lead a month to AI search. She told me that on the record. Then she spent the rest of our conversation showing me why that number says almost nothing about the channel, and what she reads in its place.
Every B2B team is sitting on email addresses nobody has identified: webinar registrants, event signups, gated downloads, trial requests and CRM records with an email and nothing else. FullEnrich's reverse email lookup turns one email, personal or work, into the person and company behind it, including name, title, seniority, company, headcount and LinkedIn, for 1 credit per match and nothing on a miss.
If you want to build your own waterfall with custom logic, Clay is the workshop for that, and FullEnrich runs inside it as a provider. If you just want the answer, FullEnrich runs the waterfall across 20+ providers for you, with paid plans from $29 a month. Try it on your last webinar list with www.fullenrich.com
Brittany is the Co-Founder and CEO of Soarion Digital. I met her on a CMO Huddles road trip through Texas, where she analyzed the website of every marketing leader in the room for AI search performance, city after city. There were some big logos in those rooms, and the results surprised me. So I asked her onto the GTM AI Podcast to show me how she thinks. We went 36 minutes, including a live look at Soarion OS, the agent platform her team runs client work on. Here’s what actually matters:
1) AI search is a new channel. Run it like one.
When Facebook ads showed up, nobody ran them like Google Ads. New platform, new behavior, new playbook. Brittany wants AI search treated the same way. “Call it AEO, AI search, AIO, whichever acronym,” she told me. “People searching with AI is a new channel.”
The behavior gap is her whole case:
Old search: a few keywords, a scan of the blue links, a click.
AI search: “I’m gonna dump my life story in there. I’m gonna give you 2, 3, 4 sentences. I’m gonna tell you my budget. I’m gonna tell you my pain points.”
Where it starts: earlier in the decision, before the buyer knows your name. “They don’t even know who you are.”
The market data backs her up. G2 surveyed 1,076 B2B software buyers in March 2026, and 51% now start their research in an AI chatbot more often than in Google. In G2’s 2025 report that number was 29%. A channel that size gets its own strategy, its own owner and its own scoreboard.
2) Click attribution undercounts AI search. Read four signals together.
I asked Brittany how much business Soarion gets from AI search. Her answer: “I can attribute one lead to AI search a month.”
Think about what has to happen for that lead to get counted. The buyer sees you in the answer, clicks the citation, lands on your site and fills out the form, with the referral source surviving the whole trip. “It breaks,” she said.
Real buyers do something simpler. They see your name in ChatGPT, open Google and type it in. “We’re creatures of habit.” Your CRM records organic search. AI did the work and Google got the credit.
So Brittany reads four signals as one picture:
Presence: how often your brand is mentioned across the prompts you track
Citations: how often AI links to your site as its source
Branded search: how many people search for you by name
Referral source: the traffic and leads that do arrive tagged from an AI platform
Soarion runs no paid ads right now. Organic, AI and events carry the pipeline. So when presence and citations climb and branded search and leads climb with them, the read is clean. My add for teams that run paid: mark every budget change on the same chart, or AI search gets credit for a lift your ad spend bought.
3) Your largest audience is already agents.
Soarion tracks agent traffic on its own website. The ratio Brittany gave me: “20 to one, 20 agents to one human.” Those agents index pages, retrieve answers and complete tasks for the people who sent them.
She lives on the buyer side of this too. The night before we recorded, she told Meta’s Muse agent to buy tickets for 10 a.m. Sunday. It did. She never visited the website.
Now run that through your own funnel. An agent shows up to compare vendors or start a signup for a buyer, and your site blocks it or hides what it needs. Brittany’s verdict: “Game over. The agent’s gonna go, I can’t go there, and then it’s gonna go to your competitor and move on.”
She expects the web to split in two: “human sites and human content, and AI sites and AI content.” Her team is testing an agent-only site right now, because “humans want different content than AI wants.”
4) Become the source before you become the recommendation.
I asked which comes first: AI citing your content or AI recommending your company. “Education all the way.”
Buyers now start their research inside AI “instead of talking to a salesperson, or talking to a solutions consultant.” That early research used to happen where you had no view of it. Now the brands that answer the early questions earn the citations. “Citations are becoming gold,” Brittany said, because the link means AI saw you as the credible source.
How to work it:
List the pain-point questions your buyer asks before they know your category exists
Publish a clear answer to each one on your own site
Track those prompts for presence and citations, next to your branded prompts
Most B2B teams starved this content for years. Her line: “You’ve ignored it for a while, but actually that’s where people are starting, and you can be that source.”
5) Fresh beats loud.
Brittany shared the latest stat she had seen: 50% of the content LLMs cite is less than 13 weeks old. One quarter. That is the clock your content runs on.
It sounds like a case for flooding the zone. She has watched teams try it: several content pieces a day, press releases, Reddit threads. “You’ll get your peak, and then you’re gonna plummet, because it starts to catch on.”
What holds up:
A steady cadence. I pitched her a plan on the show: two or three research-backed reports over six months, plus a product launch, released weekly with one consistent core message. Her reply: “I’d say that’s the perfect strategy.”
Credible third-party sites. Reddit had its hype, then one model changed its mind and the tactic died. “One thing that has stayed consistent is PR, or those credible third-party sites.”
One narrative. In a large company every product line tells a slightly different story, and AI reads all of them. Her fix: “Just start with one product line,” and put comms and marketing at the same table.
6) Strategy first: 2 to 3 clusters, 1 or 2 personas, the right platform.
Visibility tools such as Profound and Scrunch show you where you appear. Soarion OS pulls that data in and treats it as one input. “The AI visibility is really important,” Brittany said. “But it’s one tiny piece of the whole pie.”
The standard AEO pitch skips the rest of the pie: here is a prompt, we will get you to 80% presence on it. Her team is “very against that approach.” Her test: “Does it actually matter to your business, and is it actually going to drive results?”
She caught herself doing it once. She saw a presence gap and rushed to write for a single prompt. Then she stopped: “What are you doing, Brittany?” Her rule now: “Let’s not forget to be marketers.”
How Soarion builds it:
Clusters over single prompts. A cluster is a hub with spokes: the prompts you track, the content topics, your old SEO keywords and the themes in your industry.
2 to 3 clusters at a time. Everything else waits.
1 or 2 personas, even when you sell to the whole C-suite.
The persona picks the platform. “If you’re selling into finance, do you think they’re using Grok? No, they’re probably in Copilot. So I need to optimize for Copilot, not Grok.”
The proof is a client story. A B2B company was down to about 10 leads a month. “AI couldn’t even get to their site.” Soarion fixed the technical layer, picked the two services that mattered most and built the content around them. Within 90 days that client was up to 40 leads.
The tactical shift:
Brittany closed with the line I keep repeating: “It doesn’t matter what the size of your company is, we’re all at the same starting line.” AI search reset the board. Big brand or six-person startup, everyone has to rebuild the strategy for this channel, and measurement is the part everyone is still fighting with. The team that builds its scoreboard this quarter is ahead of almost everybody.
What to do this week:
Run the baseline. Ask 10 buyer questions in ChatGPT, Gemini, Claude and Perplexity. Record who gets named and who gets cited. Brittany’s team will also run it for you: her AI Search Presence Report is free and arrives in two business days.
Build the four-signal page. Presence, citations, branded search and AI referral traffic in one place, reviewed monthly.
Check your agent ratio. Ask your web team for last month’s bot report. Compare agent requests with human sessions.
Pick 2 clusters and 1 persona. Write the first three upper-funnel answers before you touch anything else.
Everyone is back at the starting line. Start running.
Coach K
Go deeper
The 13-Week AI Search Playbook, my field guide from this conversation below
Brittany’s free AI Search Presence Report: https://soariondigital.com/ai-presence-report
Brittany in writing, “AI Search Is a New Channel”: https://soariondigital.com/blog/ai-search-new-channel/
G2’s report, The Answer Economy (the 51% stat): https://company.g2.com/news/g2-research-the-answer-economy
Connect with Brittany: https://www.linkedin.com/in/brittanytrafis
13 Week AEO Strategy plan
AI search is a new channel. Brittany says it plainly: “Call it AEO, AI search, AIO, whichever acronym. People searching with AI is a new channel.”
Her analogy: when Facebook ads arrived, nobody ran them like Google Ads. New platform, new behavior, new playbook.
How a buyer used to search
Typed a few keywords
Scanned the blue links
Clicked through to a website
How a buyer searches with AI
Writes a paragraph. “I’m gonna dump my life story in there. I’m gonna give you 2, 3, 4 sentences. I’m gonna tell you my budget. I’m gonna tell you my pain points.”
Starts earlier in the decision. “They don’t even know who you are.”
Gets a written answer that names a few companies and cites a few sources
Sends an agent to do the legwork. Brittany told Meta’s Muse agent to buy tickets for 10 a.m. Sunday. It did. She never visited the website.
How big the shift is
G2 surveyed 1,076 B2B software buyers in March 2026. 51% now start their research in an AI chatbot more often than in Google, up from 29% in G2’s 2025 report. 69% chose a different vendor than they first planned because the chatbot recommended it.
Who is visiting your site
Soarion tracks agent traffic on its own website. The ratio Brittany gave me: 20 agents for every human. Those agents index pages, retrieve answers and complete tasks for the people who sent them.
The scoreboard: four signals, read together
Set this up before you change anything. You need the before picture.
Signal 1: Presence
What it is: how often your brand is mentioned across the prompts you track
Where to get it: a visibility platform such as Profound or Scrunch, or a manual run of your prompt list once a month
Signal 2: Citations
What it is: how often AI links to your website as the source of its answer
Why it matters: “Citations are becoming gold,” Brittany told me. The link means AI saw you as the credible source.
Where to get it: the same platform or the same manual run. Record every linked source, yours and your competitors’.
Signal 3: Branded search
What it is: how many people search for your company by name
Why it matters: this is where AI-influenced buyers surface. They see your name in an answer and go type it into Google. “We’re creatures of habit.”
Where to get it: Google Search Console, filtered to queries that contain your brand name
Signal 4: Referral source
What it is: the sessions and leads that arrive tagged from an AI platform
Where to get it: your web analytics
What to expect: a small number. Brittany can attribute one lead a month this way.
How to read them. Brittany’s words: “You really have to take this mix of the four of them together to be able to see if you’re having success with it.”
My read on the order: presence and citations move first. Branded search and total leads follow. When all four climb together, the channel is working.
Two adds from me:
Put a “How did you hear about us?” field on your form. Buyers will type “ChatGPT” when your analytics cannot see it.
If you run paid, mark every budget change on the same chart. Soarion runs no paid ads right now, which is why Brittany’s read is clean.
Week 0: Run the baseline
One hour.
Write 10 questions your buyer asks. Five that name your category and five that only describe the pain. Write them the way a person talks: three or four sentences, with budget and context.
Ask each one in ChatGPT, Gemini, Claude and Perplexity. Add Copilot if your buyer lives in it. Brittany’s example is finance.
For every answer, record three things: were you named, were you cited, and which competitors and sources showed up in your place.
Pull your branded search trend and your AI referral sessions for the last 90 days.
One caution from the episode: “If you and I did the same search right now, we’d both get different answers.” Treat a manual run as a sample. Run it again next month with the same questions.
Done when: one page shows your four signals with today’s date on it.
The shortcut. Brittany’s team runs this for free. The AI Search Presence Report shows which buyer questions name your competitors and leave you out, how often each competitor is named and cited across ChatGPT, Gemini, Claude and Perplexity, and where to focus first. It arrives in two business days: https://soariondigital.com/ai-presence-report
Weeks 1 to 2: Make your site readable
This comes first for a reason. Brittany’s client story started here: “AI couldn’t even get to their site, and didn’t even know that they were there.” No amount of content fixes that.
Her warning for everyone else: “If your site’s not ready for that. Game over. The agent’s gonna go, I can’t go there, and then it’s gonna go to your competitor and move on.”
Soarion runs a full technical audit for this. On their site the technical stage covers indexability, page errors and schema. Below are four checks you can run yourself this week. These four are mine.
1- Check the door. Open yoursite.com/robots.txt and the bot settings in your CDN or firewall. Confirm that the AI crawlers you want are allowed in. Names to look for: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot and Google-Extended.
Done when: you can say which AI agents are allowed and which are blocked, and somebody chose that on purpose.
2- Check what an agent sees. Open your five most important pages with JavaScript turned off. Whatever disappears is invisible to any agent that reads the raw page.
Done when: your product, pricing and proof content still show up with JavaScript off.
3- Count your agents. Ask your web team for last month’s bot report from your CDN or server logs. Compare AI agent requests with human sessions.
Done when: you know your own ratio. Brittany’s is 20 to 1.
4- Send an agent shopping. Give an AI agent your main conversion as a task: “Go to [your site], find the pricing and get to the demo request form.” Watch where it stalls.
Done when: the agent reaches your form without help, or you have a list of what stopped it.
Week 3: Set the strategy
This is where most AEO programs go sideways. The standard pitch in the industry: here is a prompt, we will get you to 80% presence on it. Brittany’s team is “very against that approach.” Her test: “Does it actually matter to your business, and is it actually going to drive results?”
She told me she caught herself doing it. She saw a presence gap and rushed to write content for one prompt. Then she stopped: “What are you doing, Brittany?” Her rule: “Let’s not forget to be marketers.”
Soarion pulls more than 10 data sources into a strategy. You can start with four decisions.
Decision 1: Your clusters. Pick 2 to 3.
A cluster is a hub with spokes. The hub is a topic you need to own. Soarion’s own example is “AI search optimization.” The spokes are everything that hangs off it:
The prompts you want to show up for
The content topics that answer them
Your old-school SEO keywords
The themes your industry is talking about
Soarion works only 2 to 3 clusters at a time. For the client in the 90-day story, the focus was the two services that mattered most.
Decision 2: Your personas. Pick 1 or 2.
Even when you sell to the whole C-suite. For each one, answer Brittany’s questions. What are their behaviors? Are they searching in AI? What are they asking?
Decision 3: Your platform.
The persona picks it. “If you’re selling into finance, do you think they’re using Grok? No, they’re probably in Copilot. So I need to optimize for Copilot, not Grok.”
Decision 4: Your influencers.
The people around your buyer ask the follow-up questions. Brittany’s example: someone hears that the CMO wants to do this, then asks AI a more specific question about the vendor being recommended. Those questions need lower-funnel content.
Done when: one page lists your 2 to 3 clusters, your 1 or 2 personas, the platform each persona uses and the prompts you will track for each cluster.
Weeks 4 to 13: Publish on the clock
Three rules from Brittany, then a cadence from me.
Rule 1: Education first.
I asked which comes first, AI citing you or AI recommending you. “Education all the way.” Buyers now start their research in AI “instead of talking to a salesperson, or talking to a solutions consultant.” So write the early answers: what the problem is, what the options are and how to choose. Her line: “You’ve ignored it for a while, but actually that’s where people are starting, and you can be that source.”
Rule 2: Fresh, every week.
Half of what LLMs cite is under 13 weeks old, per the stat Brittany shared. Her read: put out content continuously “so that it is new, is refreshed, or optimized for the AI to pick it up.” Refreshing a strong page counts.
Rule 3: No slop.
Some teams publish several pieces a day, plus press releases and Reddit threads. It works for a moment. “You’ll get your peak, and then you’re gonna plummet, because it starts to catch on.”
The cadence (mine):
One new answer page per cluster every week
One refresh of an existing page every week
One third-party placement every month: PR, a credible industry site or a research release
On third-party sites. Brittany has watched tactics come and go. Reddit had its hype, then one model changed its mind and the tactic died. “One thing that has stayed consistent is PR, or those credible third-party sites.”
I pitched her a release plan on the show: two or three research-backed reports over six months, plus a product launch, released weekly with one consistent core message. Her reply: “I’d say that’s the perfect strategy.”
If you work in a large company. Every product line tells a slightly different story, and AI reads all of them. Brittany’s fix: “Just start with one product line,” and put comms and marketing at the same table.
Done when, every week: one new answer is live, one page is refreshed and every piece maps to a prompt in one of your clusters.
Week 13: Read the scoreboard
Pull the same four signals you recorded in week 0.
Presence up on your cluster prompts?
Citations up on those prompts?
Branded search up?
More AI-referred sessions and more total leads?
My read on the three outcomes:
All four up. Keep the cadence and add the next cluster.
Presence and citations up, leads flat. AI is using your content and buyers are still choosing someone else. Write the lower-funnel answers from Decision 4.
Nothing moved. Go back to weeks 1 to 2. An agent that cannot read your site cannot cite it.
Then start the clock again. Your week 4 pages are nine weeks old now.
Six ways teams lose this channel
Running AEO like SEO with new keywords
Judging the channel on click attribution alone
Chasing 80% presence on a prompt that does not matter to the business
Publishing volume in place of answers
Optimizing for a platform their buyer does not use
Letting every product line tell a different story
Where this is heading
Brittany expects the web to split: “I fully subscribe to the vision that we’re going to have human sites and human content, and AI sites and AI content in the future.”
Her team is experimenting with it now. An agent-only site, written and managed by AI for agents, next to a site built for people. She expects the routing to be automatic. The site will tell whether a visitor is an AI and send it to the right place.
My take: you do not need two websites this quarter. You do need to know your agent ratio and whether an agent can get through your site. Those are checks 3 and 4 above.
Quick answers
Is AEO the same as SEO? AI search is a new channel with different buyer behavior. Buyers write full paragraphs, start earlier in the decision and can finish a task without visiting your website.
How do you measure AEO? Read four signals together: presence, citations, branded search and referral source. Click attribution alone undercounts the channel.
How long does AEO take to show results? One data point from the episode: a B2B client went from about 10 leads a month to 40 within 90 days, after technical fixes and focused content.
Which AI platform should I optimize for? The one your buyer uses. Brittany’s example: finance buyers are in Copilot.
Does publishing more content help? Fresh content helps. Half of what LLMs cite is under 13 weeks old, per the stat Brittany shared. Volume for its own sake gets a peak and then a plummet.
What should I fix first? Access. If AI cannot reach and read your site, content will not save you.
Does PR help AEO? Yes. Brittany says PR and credible third-party sites have stayed consistent as sources while other tactics came and went.
Every link in one place
Brittany’s free AI Search Presence Report: https://soariondigital.com/ai-presence-report
Brittany in writing, “AI Search Is a New Channel”: https://soariondigital.com/blog/ai-search-new-channel/
How Soarion runs its five stages: https://soariondigital.com/how-it-works
Brittany Trafis: https://www.linkedin.com/in/brittanytrafis
Soarion Digital:
https://soariondigital.com
G2’s report, The Answer Economy: https://company.g2.com/news/g2-research-the-answer-economy
My challenge to you, this week
Block one hour. Write your 10 buyer questions, run them through four AI platforms and record who gets named and who gets cited. Then put your four signals on one page. Done by Friday. mk? mk.
If you do this, three things happen:
You stop arguing about whether AI search works and start reading a scoreboard.
You find out who AI recommends when your buyer describes the problem you solve.
You walk into next year’s planning with a baseline your competitors do not have.

